Reuse of external metabolomic data from human cohort studies
Résumé
In recent years, the generation of metabolomic data from epidemiological studies has hugely increased, as metabolomics is now recognized as a powerful tool for a better understanding of biological systems and discovering new associated biomarkers. In this field, data sharing is particularly essential to accurately assess relationships between specific exposures and health-related outcomes. However, data deposition requirements specific to metabolomic epidemiology studies are challenging due to the complexities of large-scale demographic, covariate, and phenotypic data required for epidemiological modelling (Lasky-Su et al., 2021). In this work, we provide an overview of the state of human cohort studies data deposition in two public metabolomic data repositories, namely Metabolomics Workbench and MetaboLights, with a perspective of data reuse, in order to encourage data repository, and make suggestions to improve their adoption. To fulfil this objective, we focused on a particular example, the metabolic syndrome (MetS), which is currently a critical and important public health issue due to its increasing prevalence. Recognised as a progressive physio-pathological state, MetS is one of the components leading to pre-diabetes, type 2 diabetes (T2D) and cardiovascular diseases. First, studies and associated data, related to MetS and T2D, were extracted from the two data repositories. An assessment of their availability and accessibility was performed in comparison to the published data. Then, in a second step, the reuse of these external metabolomic data from human cohort studies were mined and compared to those obtained from published studies on MetS: a systematic review (Monnerie et al., 2020) and an original study (Comte et al., 2021). We evaluated the descriptive metadata of the subjects and metabolites and identify critical points constraining the exploitation and new use of the data, such as the misfiling of different information, and/or the inaccessibility of raw data and metadata. In an epidemiological context, we made a proof of concept of a positive and successful reuse of these metabolomic data. Indeed, results obtained on data reuse for metabolite annotation, systematic review or meta-analysis allowed us to make encouraging recommendations for data sharing enabling full exploitation of the wealth of these data in this field.
